Michael Burry Raises Concerns Over AI’s Impact on Accounting

Michael Burry, renowned investor and prominent figure from “The Big Short,” recently expressed concerns regarding the impact of artificial intelligence (AI) on accounting standards. In a post on X, Burry criticized how major companies like Oracle and Meta are managing the depreciation of their graphic processing units (GPUs).

Concerns Over Depreciation in Accounting

Burry argues that companies are understating depreciation by artificially extending the useful life of their assets. This practice can inflate earnings, which he labels a prevalent form of financial misconduct. He highlighted that significant capital expenditure (capex) on Nvidia chips and servers should not lead to an extended useful life of these assets.

Understanding Capital Expenditure and Depreciation

The ongoing boom in capital expenditure has substantially contributed to rising profits in the S&P 500. Companies investing heavily in data centers do not report these costs immediately but rather spread them out over time through depreciation. Consequently, this spending boosts revenue for other firms, creating a significant short-term gain for Corporate America.

• Estimated useful life of AI servers: 5 to 6 years
• Economic utility of AI equipment: Varies based on usage

Burry emphasizes the disparity between the depreciation schedules for servers and the product cycle for cutting-edge chips. He likens the situation to sports, where a player may still contribute value even while their peak performance has passed.

Challenges and Variability in Chip Usage

The distinctions in workloads for GPUs are crucial. Training GPUs require more resources than those used for inference, leading to varying longevity in performance. Technology companies contend that repurposing chips from training to inference extends their utility, aligning with new product releases.

The energy efficiency of newer models can also influence the perceived obsolescence of older models. For example, rental rates for Nvidia’s Hopper and Ampere GPUs have shown a downward trend, raising questions about the long-term viability of older technology.

Market Responses and Consumer Demand

Recent earnings calls from industry leaders reveal mixed insights. Notably, Nebius reported strong demand for upgraded GPU contracts. CoreWeave indicated high customer retention rates and satisfaction, citing futures contracts showing consistent pricing.

• Pleasing developments for older chips: Oracle’s GPU business generates notable revenue from older generations of Nvidia chips.
• Market resilience: Evidence suggests businesses remain willing to invest in non-flagship AI computing capabilities.

The Future of GPU Longevity

Despite the robust performance of existing models, uncertainties remain about future chips like Nvidia’s Blackwell. Technical issues can potentially alter their longevity and market performance. Burry’s perspective raises critical questions about how effectively companies could sustain cash flows from these investments.

In conclusion, while Burry’s critique opens a vital dialogue about AI’s impact on accounting, the broader implications of depreciation are still unfolding. Continued scrutiny on corporate practices will be essential as the technology sector evolves.